Merge feature/ansi-sanitization: hardened ANSI regex with OSC/CSI/C1 support

- Combined hardened ANSI_ESCAPE_PATTERN from ansi-sanitization branch
- Preserved content-block array rendering from content-block-array-rendering branch
- Both features now work together: arrays checked first, then ANSI stripped

Author: ShlomoStept

src/claude_code_transcripts/__init__.py

@@ -53,18 +53,18 @@ def get_template(name):
 )
 
 # Regex to strip ANSI escape sequences from terminal output
-ANSI_ESCAPE_PATTERN = re.compile(r"\x1b\[[0-9;]*[a-zA-Z]")
+ANSI_ESCAPE_PATTERN = re.compile(
+    r"""
+    \x1b(?:\].*?(?:\x07|\x1b\\)  # OSC sequences
+    |\[[0-?]*[ -/]*[@-~]         # CSI sequences
+    |[@-Z\\-_])                  # 7-bit C1 control codes
+    """,
+    re.VERBOSE | re.DOTALL,
+)
 
 
 def strip_ansi(text):
-    """Strip ANSI escape sequences from terminal output.
-
-    Args:
-        text: String that may contain ANSI escape codes.
-
-    Returns:
-        The text with all ANSI escape sequences removed.
-    """
+    """Strip ANSI escape sequences from terminal output."""
     if not text:
         return text
     return ANSI_ESCAPE_PATTERN.sub("", text)
@@ -2246,4 +2246,5 @@ def on_progress(project_name, session_name, current, total):
 
 
 def main():
+    # print("RUNNING LOCAL VERSION!!")
     cli()

tests/test_generate_html.py

@@ -18,6 +18,7 @@
     render_edit_tool,
     render_bash_tool,
     render_content_block,
+    strip_ansi,
     analyze_conversation,
     format_tool_stats,
     is_tool_result_message,
@@ -284,6 +285,21 @@ def test_tool_result_error(self, snapshot_html):
         result = render_content_block(block)
         assert result == snapshot_html
 
+    def test_tool_result_with_ansi_codes(self):
+        """Test that ANSI escape codes are stripped from tool results."""
+        block = {
+            "type": "tool_result",
+            "content": "\x1b[38;2;166;172;186mTests passed:\x1b[0m \x1b[32m✓\x1b[0m All 5 tests passed\n\x1b[1;31mError:\x1b[0m None",
+            "is_error": False,
+        }
+        result = render_content_block(block)
+        assert "\x1b[" not in result
+        assert "[38;2;" not in result
+        assert "[32m" not in result
+        assert "[0m" not in result
+        assert "Tests passed:" in result
+        assert "All 5 tests passed" in result
+
     def test_tool_result_with_commit(self, snapshot_html):
         """Test tool result with git commit output."""
         # Need to set the global _github_repo for commit link rendering
@@ -369,6 +385,21 @@ def test_tool_result_content_block_array_with_tool_use(self, snapshot_html):
         assert result == snapshot_html
 
 
+class TestStripAnsi:
+    """Tests for ANSI escape stripping."""
+
+    def test_strips_csi_sequences(self):
+        text = "start\x1b[?25hend\x1b[2Jdone"
+        assert strip_ansi(text) == "startenddone"
+
+    def test_strips_osc_sequences(self):
+        text = "title\x1b]0;My Title\x07end"
+        assert strip_ansi(text) == "titleend"
+
+    def test_strips_osc_st_terminator(self):
+        text = "name\x1b]0;Title\x1b\\end"
+        assert strip_ansi(text) == "nameend"
+
 class TestAnalyzeConversation:
     """Tests for conversation analysis."""